Summary: A new longitudinal study overturns a long-standing view of human multitasking by showing how repetitive practice physically remodels brain circuits. The research demonstrates that with extensive training, complex tasks can be automated by migrating out of the prefrontal “bottleneck” into temporal brain regions, freeing frontal networks to handle additional operations in parallel.
Using a combination of functional MRI (fMRI) and EEG, the investigators tracked how task representations moved in the brain after tens of thousands of practice trials. Their results show that continued training causes category-selective circuitry to emerge in the temporal cortex, effectively bypassing executive regions and creating automated pathways that support simultaneous task performance.
Key Facts
- The Frontal Bottleneck Overpass: Early in learning, the prefrontal cortex (pFC)—the brain’s executive center—manages demanding tasks and typically limits attention to one complex activity at a time.
- The Temporal Offloading Discovery: After weeks of extensive practice, the neural representation of the trained task shifts into the ventral occipito-temporal cortex (vOTC) and adjacent temporal areas, regions specialized for object recognition and memory encoding.
- The 30,000-Trial Longitudinal Audit: Study participants completed more than 30,000 image-sorting trials over a 5–10 week period using a mobile app. Researchers recorded brain activity and connectivity both before initial learning and after participants reached high levels of expertise.
- Dismantling the Task-Switching Myth: Rather than relying on rapid switching between tasks, the brain can construct distinct neural circuits through practice so that two tasks can run concurrently without vying for the same frontal resources.
- The Unlearning and Compulsion Metric: Because automated behaviors migrate into circuits that bypass conscious executive control, this work clarifies why some compulsive habits are resistant to simple cognitive strategies and points to anatomical targets for interventions.
- The Human Continuous Learning Blueprint: Offloading routine skills to temporal regions leaves the pFC available to recombine prior knowledge as modular building blocks for new learning, a mechanism that helps explain human flexibility that current artificial intelligence systems still struggle to match.
- Circuit Compatibility Horizons: Authors highlight next steps: identifying the signals that prompt neural migration and mapping the boundaries of parallel processing. Tasks that depend on the same sensory or motor mechanics may remain incompatible for safe simultaneous performance.
Source: Georgetown University
New research from Georgetown scientists reveals how the brain rewires itself to automate complex skills, offering evidence that true multitasking is achievable under the right conditions.
Beyond reassuring people that learning can enable safe parallel performance of some tasks, these findings have implications for designing artificial systems capable of continuous learning and for clinical strategies aimed at breaking deeply ingrained habits.
“We have another stepping stone in our understanding of how the brain learns,” said senior author Maximilian Riesenhuber, PhD, professor of neuroscience at Georgetown University School of Medicine and co-director of the Center for Neuroengineering. “The encouraging part is that you really can learn to multitask. There is a way to remodel your brain architecture and use other regions for automated processing.”
Earlier work focused largely on early learning stages, but the long-term evolution of task circuitry has been harder to observe. To address this, researchers taught volunteers to categorize morphed car images, training them to detect subtle differences. Participants played a sorting game on their phones and completed more than 30,000 trials in total across 5–10 weeks while researchers collected fMRI and EEG data before and after extensive practice.
Initial scans showed the categorization task engaged the prefrontal cortex, consistent with the pFC’s role in executive control. After extensive training, however, category-selective responses emerged in the ventral occipito-temporal cortex and other temporal regions. Those temporal representations had reduced functional connectivity with the pFC and stronger links to motor output areas, indicating a rerouted pathway from perception to action that bypassed frontal processing.
“Previous studies noted that experienced observers can show temporal-lobe responses to specific categories—birds, cars or even fictional characters—but most of those studies looked only after expertise developed,” said first author Patrick Cox, PhD. “Our longitudinal design lets us see that extensive practice actually creates a category-selective area that was not present before training.”
This remodeling matters for performance: the more the trained task was offloaded from the frontal bottleneck, the better participants performed on a concurrent task. In short, automated temporal circuitry increases automaticity and frees executive resources for other demands.
The study also sheds light on why certain habits are difficult to break. When behaviors run on circuits outside conscious executive control, simple strategies like “think of something else” often fail. Understanding the anatomical location of those circuits is a first step toward more targeted therapies for compulsive behaviors and addiction.
For artificial intelligence, the findings point to a structural principle humans exploit: moving stable, practiced knowledge into a specialized store so frontal-like systems remain available to compose and learn new tasks without overwriting old ones. Current AI approaches generally lack this kind of flexible scaffolding, which helps explain limits on continuous learning.
Next research priorities include identifying the neural signals that trigger migration of task representations and determining which kinds of tasks can be trained to run in parallel. As Cox notes, compatibility depends on whether independent neural circuits can be formed; tasks that share critical sensory or motor channels—such as texting and driving—remain unsafe to combine.
Funding: The study received support from the National Science Foundation (BCS-1232530), the ARCS Foundation, and the Army Research Laboratory (W911NF-24-1-0097). The authors report no personal financial interests related to the study.
Key Questions Answered:
A: Extensive practice builds category-selective circuitry where none existed before. The study shows that tens of thousands of trials can rewire processing so a practiced task migrates from the prefrontal cortex to the temporal lobe, forming an automated pathway.
A: Deeply learned behaviors move into regions that bypass conscious executive control, so simply redirecting attention rarely interrupts the underlying circuit driving the habit.
A: The brain’s strategy of offloading stable skills to specialized regions while keeping executive capacity free allows continuous learning without erasing prior knowledge—a structural approach AI could emulate to improve lifelong learning.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The journal paper was reviewed in full.
- Additional context was added by staff to summarize implications and next steps.
About this neuroscience and neurotech research news
Author: Karen Teber
Source: Georgetown University
Contact: Karen Teber – Georgetown University
Image: Image credit: Neuroscience News
Original Research: Closed access. “Extensive Experience Remodels Neural Task Circuitry to Escape the Frontal Bottleneck and Increase Automaticity of Categorization” by Patrick H. Cox, Clara A. Scholl, Marissa L. Laws, Nelson E. Jaimes, Xiong Jiang, and Maximilian Riesenhuber. Journal of Cognitive Neuroscience. DOI: 10.1162/JOCN.a.2618
Abstract
Extensive Experience Remodels Neural Task Circuitry to Escape the Frontal Bottleneck and Increase Automaticity of Categorization
Object category learning is a core cognitive function. Most laboratory studies last only a few hours and show stronger shape tuning in visual cortex along with task-dependent responses in prefrontal cortex. Other work has identified a “frontal bottleneck” that constrains multitasking. Real-world expertise, however, typically develops over weeks, months or years and can produce qualitative shifts toward automatic processing.
To test whether extensive training causes a spatio-temporal reorganization of categorization circuitry, participants completed more than 30,000 categorization trials across 5–10 weeks using a mobile app. fMRI and EEG were used to measure neural responses after initial learning (~4 hours) and after extensive practice (~16 additional hours). Converging evidence showed that visual regions in ventral occipito-temporal cortex, initially shape-selective, developed category-selective responses after prolonged training. These regions reduced connectivity with prefrontal cortex and increased connectivity with motor output areas, consistent with decision processes occurring outside the frontal bottleneck. Importantly, reduced vOTC–pFC coupling correlated with better dual-task categorization performance, demonstrating increased automaticity. These results indicate that prolonged experience transforms categorization from an attentionally controlled process into a streamlined, automatic one.